We consistently analyse for the first time the impact of survey depth and spatial resolution on the most used morphological parameters for classifying galaxies through non-parametric methods: Abraham and Conselice-Bershady concentration indices, Gini, M20 moment of light, asymmetry, and smoothness. Three different non-local data sets are used, Advanced Large Homogeneous Area Medium Band Redshift Astronomical (ALHAMBRA) and Subaru/XMM-Newton Deep Survey (SXDS, examples of deep ground-based surveys), and Cosmos Evolution Survey (COSMOS, deep space-based survey). We used a sample of 3000 local, visually classified galaxies, measuring their morphological parameters at their real redshifts (z ˜ 0). Then we simulated them to match the redshift and magnitude distributions of galaxies in the non-local surveys. The comparisons of the two sets allow us to put constraints on the use of each parameter for morphological classification and evaluate the effectiveness of the commonly used morphological diagnostic diagrams. All analysed parameters suffer from biases related to spatial resolution and depth, the impact of the former being much stronger. When including asymmetry and smoothness in classification diagrams, the noise effects must be taken into account carefully, especially for ground-based surveys. M20 is significantly affected, changing both the shape and range of its distribution at all brightness levels. We suggest that diagnostic diagrams based on 2-3 parameters should be avoided when classifying galaxies in ground-based surveys, independently of their brightness; for COSMOS they should be avoided for galaxies fainter than F814 = 23.0. These results can be applied directly to surveys similar to ALHAMBRA, SXDS and COSMOS, and also can serve as an upper/lower limit for shallower/deeper ones.